Assimilation of OMI Ozone for Air Quality Prediction

1
Assimilation of OMI Ozone for Air Quality
Prediction

• Brad Pierce1, Jay Al-Saadi2, Jim Szykman3, Todd
  Schaack4, P.K. Bhartia5, Anne Thompson6
• NOAA NESDIS
• NASA Langley Research Center
• US Environmental Protection Agency
• University of Wisconsin/SSEC
• NASA GSFC
• Penn State University
• Presented by Jay Al-Saadi

2
RAQMSglobal (2x2) 2006 Ozone Assimilation/Forecast
Procedure
Ecosystem/Severity Based wildfire emissions
MODIS Rapid Response Wild fire locations
RAQMS Global Met/Chem
NOAA GFS Global Met
5-day RAQMS Global Met/Chem Forecast
OMI O3 Column (OMIDAPS)
Real-time OMI Cloud-cleared column OI
assimilation conducted at 2x2 degrees for Feb
01-Oct 15, 2006. Analysis increment applied as
percentage adjustment to mixing ratio resulting
in minimal impact on troposphere.
3
Total Column Ozone Timeseries August 01- October
15, 2006
Assimilation of cloud-cleared OMI ozone column
captures observed mid-latitude temporal and
latitudinal variations. RAQMS analysis
underestimates mid-September through mid-October
(Julian day 260-290) Antarctic ozone loss by 17
4
Tropospheric Ozone Column (subtracting RAQMS
stratosphere) August 2006
Analyzed Tropospheric Ozone Column (TOC) well
correlated with OMI - RAQMSSTRAT (r0.84
globally, 0.90 CONUS). TOC median high bias of
3.6 Dobson Units over CONUS. Globally, median
bias is negative due to low tropical values.
5
Comparison of RAQMS NRT OMI analysis with IONS
ozonesondes (373 sondes, August, 2006)
PI ANNE M. THOMPSON Penn State
RAQMS NRT OMI Tropospheric Ozone Column (August
01-31, 2006)
Large high bias (50-100) in upper
troposphere/lower stratosphere
6
TexAQS 2006 RAQMS O3 vs NOAA P3 all flights
In East Texas, RAQMS analysis has approx 15 ppb
high bias through free troposphere. No median
bias near the surface although variance is
underestimated.
7
TexAQS, Aug-Sep 2006 Assessment of O3 Analysis
with Hourly Surface O3 Measurements
O3 timeseries from monitoring locations in
Houston and Dallas show that the RAQMS analysis
captures the temporal behavior of surface O3 at
these sites (1-hour and 24-hour r values above
0.8) but has a persistent high bias.
8
TexAQS, Aug-Sep 2006 Assessment of O3 Analysis
with Hourly Surface O3 Measurements
18Z Bias (model-obs) (ppbv)
18Z Correlation
National maps of time averaged site statistics
show RAQMS 18Z surface O3 analysis is high by 15
ppbv through most of the SE US and by 30 ppbv
over the Ohio River valley and Northeast (left)
but is highly correlated with measurements over
most of the Eastern US (right).
9
TexAQS, Aug-Sep 2006 Assessment of O3 Forecasts
with Hourly Surface O3 Measurements
18Z Bias (model-obs) (ppbv)
18Z Correlation
Histograms show the changes in bias and temporal
correlation through the 5-day forecast period.
There is little change in the distribution of
biases or median bias during the forecast period
(left). Temporal correlation coefficients (right)
show significant skill through 30 hours (median
correlation 0.6) and a systematic degradation at
forecast periods of 54 hours and beyond.
10
Extra slides
11
RAQMS August, 2006 Mean OMI O3 O3 Analysis
Increments (total column)
Assimilation of OMI total column O3 measurements
results in small (1-1.5) mean increases in total
column O3 over CONUS.